Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
Tools/GitHubGitHub/abraxas/loom-ws-hitl
Authentication & AuthorizationVulnerability AnalysisWeb SecurityPenetration TestingAPI Security
GitHubabraxas/loom-ws-hitl

loom-ws-hitl

A Loom user who can invoke an agent over the websocket skips the approval policies that an HTTP invoke would send. An anonymous websocket is still closed. v1.9.0 and v1.10.3. Medium 6.5.

View Repository
2 days agoNot yet reviewed
Share

Abraxas Labs - loom-ws-hitl

loom-ws-hitl

Someone who can already invoke an agent over the websocket skips the approval list that an HTTP invoke would attach. The socket never loads ApprovalPolicy. Tools that were supposed to wait for a person do not get that list.

An anonymous socket is still closed. This lab did not run a tool through to a side effect. It showed the HTTP payload carrying one policy and the socket payload carrying none. The score treats the skipped gate as high integrity because that gate is what the product uses to stop the tool. The oracle itself is the missing key.

Class: Missing authorization check on the invoke socket Reach: Remote. In-group invoke on a custom-code agent. Not anonymous.

Loom v1.9.0 a3050669 and v1.10.3 cc10ab0

IDno CVE yet
ClassWebsocket invoke omits approval_policies
ReachRemote, in-group invoke
CWECWE-862
CVSSMedium: 6.5 CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:N
Productawslabs/loom
Affectedv1.9.0 a3050669f5b4c83210a44ed4346f17e01641e82f and v1.10.3 cc10ab08573087fa12154f7c8181ffcd985fea88
Authfirst-frame bearer with invoke and the agent's group
LicenseGNU Affero GPL v3.0
Labloopback Docker on v1.9.0. v1.10.3 was a source check.

What an attacker can do

PrivilegeMethod / sinkImpactLab-proven or inferred
Open the socket with no tokenWS /api/agents/{id}/wsClosedLab-proven. Unauthorized, close 1008, session-start=0
Send the approval list on HTTPPOST /api/agents/{id}/invokeThe payload includes approval_policiesLab-proven. Count 1.
Invoke on the socket without that listWS /api/agents/{id}/ws after a valid first frameIntegrity. Payload keys were type, prompt, session_id, actor_id.Lab-proven. ws-approval-policies=0
Run a specific tool to completion with no approval rowthe agent processIntegrityInferred from the omit. The lab stopped at the payload.

Stays closed

AxisMethod / sinkWhy it is closed
IntegrityAnonymous invokeFirst frame is authenticated. No session starts.
IntegrityIAM-signed runtimesThe socket does not sign
IntegrityHTTP invoke of the same agentStill loads enabled policies

How I found it

HTTP invoke queries enabled approval rows and passes them through. invoke_agent_websocket does not query that table, and invoke_agent_ws has no approval_policies field. I assumed the socket had been brought up to the HTTP contract when the unauthenticated connect was closed. The auth check landed. The policy query did not.

Container on v1.9.0. The split is still there on v1.10.3.

Lab

python3 loom-ws-hitl-Abraxas-Labs.py
SUCCESS LOOM-190-WS-HITL unauth=Unauthorized session-start=0 http-approval-policies=1 ws-approval-policies=0 LOOM-190-WS-HITL-WITNESS

The fix

Load the same enabled ApprovalPolicy rows in invoke_agent_websocket and pass them into invoke_agent_ws.

References

  • awslabs/loom v1.10.3 cc10ab08573087fa12154f7c8181ffcd985fea88
  • invocations.py invoke_agent_websocket
  • agentcore.py invoke_agent_ws
  • CWE-862
  • CVSS 3.1 calculator

License

GNU Affero GPL v3.0. See LICENSE. Loopback lab only. No warranty.

Abraxas Labs [email protected]

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    867
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    544
  • 3

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    385
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    360
  • 5

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    302
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    279

Explore all tools

Browse our collection of tools

View all tools →